Resumen en inglés

Deforestation is reshaping Colombia’s landscapes—and potentially the earliest stages of human development—but evidence on neonatal health has been scarce. Forest loss can degrade air quality through biomass burning, alter vector ecology in ways that favor malaria transmission, disrupt forest-based diets and livelihoods, and raise local temperatures; together, these channels expose pregnant women to pollution, infectious disease risk, nutritional shocks, and heat stress, all of which are known determinants of fetal growth and survival. I bring these mechanisms to the Colombian frontier context by linking satellite-based tree-cover loss (Global Forest Watch, 2001–2023) to the universe of Colombian vital statistics and estimate two-way fixed-effects models with municipality×year and municipality×month fixed effects plus rich maternal covariates. Exposure is the annual deforestation rate (percent of year-2000 tree cover lost), rescaled so coefficients read per 10 percentage points. Across 13.5 million births, a 10-pp increase in deforestation is associated with tiny improvements in Apgar at 1 and 5 minutes (~0.07 points) and a modest decline in low birthweight (–0.69 p.p.), while prematurity and neonatal mortality are indistinguishable from zero; patterns are similar in frontier-focused samples. These are correlations, not causal effects. The attenuation—and in some cases reversal—of raw adverse associations once rich fixed effects are introduced suggests that environmental harms and short-run economic shifts from land-use change may offset each other in reduced form. The contribution is to identify deforestation exposure as a potential risk factor and to map where environmental and health vulnerabilities intersect. The results motivate future designs leveraging plausibly exogenous variation (e.g., enforcement or commodity shocks, rainfall) and higher-frequency pollution/heat/malaria data to isolate mechanisms and establish causal impacts.